Biometrical Journal Статистика, вероятность и неопределённость

Graphical Approaches for Multiple Comparison Procedures Using Mixed Types of Intersection Tests

Dong Xi, Ethan Brockmann
Wiley · 2026-11-20 · DOI 10.1002/bimj.70205

Аннотация

ABSTRACT The graphical approach to multiple testing provides a popular tool for designing, visualizing, and performing multiplicity adjustments in confirmatory clinical trials while controlling the familywise error rate. Graphical approaches have been extended to closed test procedures using a single type of intersection test, including weighted Bonferroni, weighted Simes, Hochberg, or weighted parametric tests. In this paper, we provide a unifying framework that accommodates mixed types of these tests within a single graphical multiple comparison procedure. Hypotheses are partitioned into prespecified groups, each group is tested by its own type of intersection test at its group‐specific significance level, and the group‐level results are combined through the closure principle. We prove that the resulting procedure controls the familywise error rate in the strong sense whenever each group‐level test is valid, without any assumption on the dependence between groups. The proposed procedure is at least as powerful as the corresponding procedure using weighted Bonferroni tests only. We illustrate the utility of this framework using a clinical trial, and demonstrate its implementation using the graphicalMCP R package.

Читать у издателя →